Related Experiment Videos
Group sequential designs using a family of type I error probability spending functions
I K Hwang1, W J Shih, J S De Cani
1Biostatistics & Research Data Systems, Merck Sharp & Dohme Research Laboratories, Rahway, NJ 07065-0914.
Statistics in Medicine
|December 1, 1990
Summary
This study introduces a flexible method for analyzing accumulating clinical trial data. It offers customized group sequential boundaries for interim analyses, improving statistical power and trial design.
Area of Science:
- Biostatistics
- Clinical Trial Design
- Statistical Methods
Background:
- Interim analyses of accumulating data are common in clinical trials.
- Existing methods for constructing group sequential boundaries have limitations.
Purpose of the Study:
- To propose a generalized family of type I error probability spending functions.
- To enable the construction of customized group sequential boundaries with unequal information time increments.
Main Methods:
- Development of a one-parameter family of spending functions.
- Generalization of existing Lan and DeMets and Kim and DeMets spending functions.
- Application to group sequential trial design with unequal information increments.
Main Results:
- A flexible and generalizable family of spending functions is proposed.
- The proposed family allows for customized group sequential boundaries.
- An example demonstrates the practical utility of the proposed method.
Conclusions:
- The proposed spending function family offers enhanced flexibility for group sequential trial design.
- This method facilitates more adaptive and efficient clinical trial monitoring.
- The approach provides a valuable tool for statisticians designing clinical trials.